Research Data Analyser

Dataset Overview

Variable Summary

Deep scan of data structure, missingness, and format errors.

Observations

Variables

Missing Cells

Duplicate Rows

'Content' shows Mean (SD) for numeric variables, or Top Category (%) for factors.

Variable Management & Modification

Create calculated fields, subset data, delete variables, or recode values. Changes apply to working dataset.

New Variable Name:

Calculation Type:

Select Variables to Delete:

Hold Ctrl/Cmd to select multiple

Warning: Deletion cannot be undone unless you reload data or click 'Reset to Original'.

Variable for Filtering:

Condition:

Current Filters Applied:

Data Preview

Before (Counts)

After (Preview)

Select Variable:

New Name:

Use descriptive names. Avoid spaces (use underscores instead).

Preview (First 10 rows)

Manage missing values (NA). You can use simple imputation (mean/mode) or advanced AI-based imputation (MICE).

Imputation
Row Deletion (Threshold)

Enable Multivariate Imputation (MICE)

Switch to 'mice' package to handle complex missing data patterns.

Select Variable to Impute:

Imputation Method:

Enter Value:

Pre-Imputation Status

Note: Standard methods update one variable at a time.

MICE Mode: Advanced imputation using chained equations.

1. Target Variables (To Impute):

Only these columns will be modified.

2. Predictor Variables (Context):

Variables used to learn the patterns (leave empty for All).

Algorithm:

Imputations:

Iterations:

Workflow Guide

Targets: The columns that have missing values you want to fix.
Predictors: Columns that provide information. MICE uses correlations in predictors to guess the targets.

Example: Use 'Age' and 'Weight' (Predictors) to fill in missing 'Height' (Target).

Select Variables to Check:

Hold Ctrl/Cmd to select multiple.

Deletion Condition

Delete row if missing count is:

Operator:

Count Limit:

When to use this tab: If any regression assumption check shows FAIL , use the tools below to transform or clean your variables, then re-run the regression with the corrected variable(s).

Create a new transformed variable from an existing one. Use for: Linearity, Normality, Homoscedasticity, Log-linearity (Cox/Binary) FAIL.

1. Select variable to transform:

2. Transformation:

3. New variable name:

Delete a variable that causes multicollinearity. Use for: Multicollinearity (VIF) FAIL.

Deletion is permanent for this session. Use Reset to Original in Dataset Overview to undo.

Select variable(s) to remove:

Hold Ctrl/Cmd to select multiple.

Exclude rows with extreme values on a selected variable. Use for: Influential Outliers FAIL.

1. Select numeric variable:

2. Exclusion method:

Merge small or sparse categories into one group. Use for: Complete Separation (Binary regression) FAIL.

1. Select categorical variable:

3. Label for merged group:

Propensity Score Matching

Create matched dataset to control for confounding variables in observational studies.

Treatment Variable (Binary):

Binary exposure/treatment variable

Covariates for Matching:

Variables to balance between treatment groups

Matching Method:

Nearest is most common

Matching Ratio:

Controls per treated case

Caliper (Optional):

Maximum propensity score difference

Covariate Balance Assessment

Download Modified Dataset

Export the current working dataset with all modifications applied.

Download as CSV

Download as Excel

Instructions: Select your analysis method, define variables, and click Run. Use the new tools to check collinearity or perform stepwise selection.

Step 1: Analysis Settings

Analysis Method:

Correction for zero events (Firth's method)

Use if you have small sample sizes or '0' events in a group (Separation). Stepwise selection is disabled for Firth.

Model Refinement

Perform Stepwise Selection (AIC)

Automatically selects the best variables using Backward/Forward AIC selection.

Click to check for multicollinearity before running the final model.

Step 2: Model Variables

Outcome Variable (Binary):

Predictor Variables (Multivariable):

Variables selected here will be included in the Multivariable columns.

Cox proportional hazards regression for time-to-event analysis. Requires time variable and event indicator (0/1).

Core Variables

Time to Event Variable:

Select numeric time variable

Event Indicator (0/1 or Binary):

Kaplan-Meier Survival Analysis

Stratification Variable:

Variable to stratify survival curves

Show 25th, 50th, 75th percentile survival

Time Points for Survival Estimates:

Comma-separated time points

Cox Regression Analysis

Model Refinement

Perform Stepwise Selection (AIC)

Automatically selects optimal predictors using bidirectional AIC selection.

Assess multicollinearity among covariates (VIF > 5 indicates concern).

Select Predictor Variables for Cox Regression:

Select variables for univariable and multivariable Cox regression

Sensitivity Analysis — Influential Observations (Cox Regression)

Identifies poorly-fitted or influential observations using either deviance residuals (|d| > threshold, default 3) or Cook's distance (> 4/n). The model is re-run after exclusion and estimates are compared side-by-side.

Settings

Detection method:

Threshold:

Show covariate values for flagged rows

Download CSV Download HTML

Develop clinical prediction scores, analyse ROC curves, and calculate diagnostic accuracy.

1. Score Development & Weighting

Outcome Variable (Binary):

Predictor Variables:

Select variables to include in the score

Variable Weighting Analysis

View coefficients to determine appropriate weights.

Logistic Regression Results (Basis for Weighting):

Assign Weights

Name of New Score Variable:

Score Created! You can now use this variable in the ROC Analysis below.

2. Clinical Risk Calculator (Simulation)

Patient Inputs

3. ROC Analysis

Select Predictor / Score to Test:

Performance

Optimal Cutoff (Youden)

4. Save & Download Data

Download the dataset including the newly created score variable.

Download as CSV

Download as Excel

5. Diagnostic Evaluation Tools

Diagnostic Accuracy Calculator

Enter values from a 2x2 contingency table.

True Positive (TP):

False Positive (FP):

False Negative (FN):

True Negative (TN):

Likelihood Ratio Calculator

Calculate post-test probability.

Pre-test Probability (%):

LR (+):

LR (-):

Enhanced Psychometric Analysis: Evaluate knowledge tests, clinical scales, or surveys with mode-specific diagnostics, manual item reversal, and comprehensive reliability/validity metrics.

Analysis Configuration

Select Analysis Mode:

Clinical Scale Mode

Analysis Type:

Maximum Score Per Item:

Enter the maximum score for EACH question/item (e.g., 5 for 0-5 scale, 4 for 0-4 scale). Total maximum will be calculated automatically.

Item Selection

Select Items (Questions/Variables):

Select 3+ numeric items for analysis.

Enable Manual Item Reversal

Item Reversal

Select items to reverse-score:

Reverse scoring: (max + min) - original_score. E.g., 5-point scale: 5→1, 4→2, 3→3.

Advanced Settings

Calculate McDonald's Omega (computationally intensive)

Factor Analysis: Number of Factors:

ICC Model:

1. Reliability & Internal Consistency

2. Overall Statistics & Score Distribution

Score Distribution

Show Individual Student Scores

3. Item Health Check

Kaiser-Meyer-Olkin (KMO) and Bartlett's Test assess whether your data is suitable for factor analysis.

Item Response Theory (IRT) curves showing the probability of correct response by ability level.

Data Import

Variable Type Editor

Report Style

Dataset Overview

Variable Summary

Variable Management & Modification

Current Filters Applied:

Recode Values

Data Preview

Before (Counts)

After (Preview)

Reshape: Wide to Long

Preview (First 10 rows)

Pre-Imputation Status

Workflow Guide

Deletion Condition

Propensity Score Matching

Covariate Balance Assessment

Download Modified Dataset

Visual Exploration

Cutpoint Analysis Tool

Step 1: Variable Selection

Analysis Summary

Method: ROC Curve Analysis

Optimal Cutpoint Results

ROC Curve

Distribution with ROC Cutpoint

Method: Maximally Selected Rank Statistics

MaxStat Cutpoint Results

Kaplan-Meier Survival Curves

Outcome Distribution by Groups

Distribution with MaxStat Cutpoint

Method: LOWESS Smoothing

Plot Settings

LOWESS Smooth Curve

Pattern Interpretation

Derivative Analysis: Rate of Change

Comparing Methods

Cutpoint Comparison Table

Visual Comparison

Inferential Statistics Calculator

Correlation Analysis

Multiple Correlation Matrix

Correlation Coefficients & P-values

Linear Regression Analysis

Model Refinement

Model Diagnostics

Repeated Measures Analysis (Linear Mixed Effects)

1. Define Data Structure

Model Specification

Regression Results

Plot Settings

Longitudinal Visualization

Step 1: Analysis Settings

Model Refinement

Step 2: Model Variables

Core Variables

Kaplan-Meier Survival Analysis

Cox Regression Analysis

Model Refinement

1. Score Development & Weighting

Variable Weighting Analysis

Assign Weights

2. Clinical Risk Calculator (Simulation)

Patient Inputs

3. ROC Analysis

Performance

Optimal Cutoff (Youden)

4. Save & Download Data

5. Diagnostic Evaluation Tools

Diagnostic Accuracy Calculator

Likelihood Ratio Calculator

Analysis Configuration

Clinical Scale Mode

Item Selection

Item Reversal

Advanced Settings

1. Reliability & Internal Consistency

2. Overall Statistics & Score Distribution